Do all the things like ++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatarSign Up
sbiewald380521dContra argument: A maintainer isn't responsible if something does not compile on rarely used operating systems.
It builds in CI and more often used Linux distributions got it to compile (or where did you think you got your binary from?) as well.
Yes, the .appimage has obviously been compiled somewhere.
But so has been MacOS...
And i get, that the sets of people managing to somehow compile it and the set of people being developers of the software aren't strictly identical.
I just am really pissed about, that after getting all that absurd dependencies together (the count of new packages just for FreeCAD was 106 and a few where so obscure, i had to dig them up in two other repos), it didn't even compile.
I actually think, it is a systemic problem. Bigger opensource projects tend to generally be absurdly hard to get to actually build outside the maintainers' machines.
If Chromium, Firefox or LibreOffice wouldn't be available with a well-maintained ebuild in the repo, there probably would be a zero chance, that i could ever hope to get them build before dieing of old age while trying.
But on the bright side: Code maintainability is of no issue if whoever comes after you can't get it to build anyways...
sbiewald380520d@Oktokolo It does not compile outside the maintainer's machine? Come on, that's why they are using CI, it is a reproducible environment for exact that reason.
Don't get me wrong, it is annoying as hell to not be able to compile some software - but I hardly see the maintainer in fault if it does not work on exotic systems I chose by myself.
Fast-Nop2948520dOSS projects should have clear documentation including dependencies, plus clear and working build instructions. Otherwise, this is an outright bug.
I actually think, that the CI makes things even worse. They probably got that working in a docker container once and then never thought about the complexity of the build process ever again.
But if anyone wants to build it from source somewhere else, they have two options:
Use their VM image or docker container containing the build environment - wich basically equals getting a blob...
Or trying to do something, wich the devs themselves maybe did only once a shitton of versions ago: Actually getting it to compile in the wild.
I am aware, that my rant is unfairly singling out a tiny team doing excellent work building a free and open parametric CAD/CAM software from scratch.
My emotional attachment to it (i even spent weeks learning how to do basic things with it despite having a pirate copy of SolidWorks lying around wich i obtained for a friend) definitely made me overreact.
But i really don't care about web browsers or office suites enough to rant about them...
Fast-Nop2948519dWith bigger projects, OSS often goes downhills. The build chain is insane so I won't actually figure out the root cause, hence no patch suggestion, and bug reports are generally ignored.
I've actually stopped filing bug reports for sizeable OSS projects unless it causes some weird workaround on my end. Then I file one, but only as reference for my own documentation. I don't expect anything to get fixed because it won't.